mirror of
https://github.com/timescale/timescaledb.git
synced 2025-05-23 06:22:03 +08:00
0b87a069e7
- Added creation_time attribute to timescaledb catalog table "chunk". Also, updated corresponding view timescaledb_information.chunks to include chunk_creation_time attribute. - A newly created chunk is assigned the creation time during chunk creation to handle new partition range for give dimension (Time/ SERIAL/BIGSERIAL/INT/...). - In case of an already existing chunk, the creation time is updated as part of running upgrade script. The current timestamp (now()) at the time of upgrade has been assigned as chunk creation time. - Similarly, downgrade script is updated to drop the attribute creation_time from catalog table "chunk". - All relevant queries/views/test output have been updated accordingly. Co-authored-by: Nikhil Sontakke <nikhil@timescale.com>
93 lines
4.8 KiB
SQL
93 lines
4.8 KiB
SQL
-- This file and its contents are licensed under the Timescale License.
|
|
-- Please see the included NOTICE for copyright information and
|
|
-- LICENSE-TIMESCALE for a copy of the license.
|
|
|
|
\c :TEST_DBNAME :ROLE_SUPERUSER
|
|
CREATE OR REPLACE FUNCTION _timescaledb_internal.test_merge_chunks_on_dimension(chunk REGCLASS, merge_chunk REGCLASS, dimension_id INTEGER)
|
|
RETURNS VOID
|
|
AS :TSL_MODULE_PATHNAME, 'ts_test_merge_chunks_on_dimension' LANGUAGE C VOLATILE;
|
|
\c :TEST_DBNAME :ROLE_DEFAULT_PERM_USER
|
|
|
|
CREATE TABLE test1 ("Time" timestamptz, i integer, value integer);
|
|
SELECT table_name FROM Create_hypertable('test1', 'Time', chunk_time_interval=> INTERVAL '1 hour');
|
|
SELECT table_name FROM add_dimension('test1', 'i', number_partitions=> 2);
|
|
ALTER TABLE test1 set (timescaledb.compress, timescaledb.compress_segmentby='i', timescaledb.compress_orderby='"Time"', timescaledb.compress_chunk_time_interval='2 hours');
|
|
|
|
-- This creates chunks 1 - 3 on first hypertable.
|
|
INSERT INTO test1 SELECT t, 1, 1.0 FROM generate_series('2018-03-02 1:00'::TIMESTAMPTZ, '2018-03-02 3:00', '1 minute') t;
|
|
-- This creates chunks 4 - 6 on first hypertable.
|
|
INSERT INTO test1 SELECT t, 2, 1.0 FROM generate_series('2018-03-02 1:00'::TIMESTAMPTZ, '2018-03-02 3:00', '1 minute') t;
|
|
|
|
CREATE TABLE test2 ("Time" timestamptz, i integer, value integer);
|
|
SELECT table_name FROM Create_hypertable('test2', 'Time', chunk_time_interval=> INTERVAL '1 hour');
|
|
|
|
-- This creates chunks 7 - 9 on second hypertable.
|
|
INSERT INTO test2 SELECT t, 1, 1.0 FROM generate_series('2018-03-02 1:00'::TIMESTAMPTZ, '2018-03-02 3:00', '1 minute') t;
|
|
|
|
SELECT id, hypertable_id, schema_name, table_name, compressed_chunk_id, dropped, status, osm_chunk FROM _timescaledb_catalog.chunk;
|
|
|
|
\set ON_ERROR_STOP 0
|
|
|
|
-- Cannot merge chunks from different hypertables
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_3_7_chunk', 1);
|
|
|
|
-- Cannot merge non-adjacent chunks
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_1_3_chunk', 1);
|
|
|
|
-- Cannot merge same chunk to itself (its not adjacent to itself).
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_1_1_chunk', 1);
|
|
|
|
-- Cannot merge chunks on with different partitioning schemas.
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_1_4_chunk', 1);
|
|
|
|
-- Cannot merge chunks on with non-existant dimension slice.
|
|
-- NOTE: we are merging the same chunk just so they have the exact same partitioning schema and we don't hit the previous test error.
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_1_1_chunk', 999);
|
|
|
|
|
|
\set ON_ERROR_STOP 1
|
|
|
|
-- Merge on open (time) dimension.
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_5_chunk','_timescaledb_internal._hyper_1_6_chunk', 1);
|
|
|
|
-- Merge on closed dimension.
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal._hyper_1_1_chunk','_timescaledb_internal._hyper_1_4_chunk', 2);
|
|
|
|
SELECT compress_chunk(i) FROM show_chunks('test1') i;
|
|
|
|
\set ON_ERROR_STOP 0
|
|
|
|
-- Cannot merge chunks internal compressed chunks, no dimensions on them.
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension('_timescaledb_internal.compress_hyper_2_10_chunk','_timescaledb_internal.compress_hyper_2_11_chunk', 1);
|
|
|
|
\set ON_ERROR_STOP 1
|
|
|
|
-- This creates more data so caggs has multiple chunks.
|
|
INSERT INTO test1 SELECT t, 1, 1.0 FROM generate_series('2018-03-02 3:00'::TIMESTAMPTZ, '2018-03-03 3:00', '1 minute') t;
|
|
|
|
CREATE MATERIALIZED VIEW test_cagg
|
|
WITH (timescaledb.continuous) AS
|
|
SELECT i,
|
|
time_bucket(INTERVAL '1 hour', "Time") AS bucket,
|
|
AVG(value)
|
|
FROM test1
|
|
GROUP BY i, bucket;
|
|
|
|
-- Merging cagg chunks should also work.
|
|
WITH adjacent_slices AS
|
|
(SELECT S1.id AS PRIMARY,
|
|
s2.id AS secondary
|
|
FROM _timescaledb_catalog.dimension_slice s2
|
|
INNER JOIN _timescaledb_catalog.dimension_slice s1 ON s1.range_end = s2.range_start
|
|
WHERE s1.dimension_id = 4
|
|
AND s2.dimension_id = 4
|
|
LIMIT 1),
|
|
chunks AS
|
|
(SELECT c1.chunk_id AS primary_chunk,
|
|
c2.chunk_id AS secondary_chunk
|
|
FROM adjacent_slices
|
|
INNER JOIN _timescaledb_catalog.chunk_constraint c1 ON c1.dimension_slice_id = adjacent_slices.primary
|
|
INNER JOIN _timescaledb_catalog.chunk_constraint c2 ON c2.dimension_slice_id = adjacent_slices.secondary)
|
|
SELECT _timescaledb_internal.test_merge_chunks_on_dimension(format('_timescaledb_internal._hyper_4_%s_chunk', chunks.primary_chunk), format('_timescaledb_internal._hyper_4_%s_chunk', chunks.secondary_chunk), 4)
|
|
FROM chunks;
|